The Key to great Improvisation is detailed Preparation

[Yora]

Browsing through the threads from recent months, one thing that was discussed briefly as being unappealing in (the common perception of) hexcrawling campaigns is the idea of GMs making shit up as they go because they don't want to bother with preparation.

I think the true value of random tables is not in generating new content, but in randomizing prepared content.

First thing: If we have to prepare content anyway, then why bother with randomizing it?

As I see it, what all RPGs really should be is a game in which the players take charge of the destiny of their characters. Things happen to the characters because the players made choices. The players have space to make decisions that have major impact on what's going to happen next, but they will also have to deal with things that don't fit into their plans, if they like it or not. This is a game after all, not collaborative story writing. Since the GM is in charge of the world, he can put anything in front of the PCs that he wants to at any time, and has the power to set up things in such ways that there's really only one thing the players can do in response to it, or that anything they do will lead to the same outcome. When that happens, players have no agency. It's railroading.
Agency is the entire point of this exercise, and players must be aware of the agency they have and genuinely feel that it is their choices that matter more than the whim of the GM. That is why GMs have to randomize at least some of the content they present to the players, and in my opinion should let the players see the randomization process.
When the party is marching through the wilderness, it's a completely difference experience when they encounter a kind and number of monsters the GM has selected, in a location that the GM has selected, at a time that the GM has selected, based on the exact knowledge of the current resources and condition of the PCs, compared to when the players run into a randomized group of monsters in a randomized location at a randomized point of time. If you get attacked by a bunch of ghouls that come out of the swamp water while you're all out of arrows, have only one healing potion, and the cleric is dead and this encounter was set up by the GM, it's easy to feel that the GM is deliberately kicking you while you are down. If that same encounter happpens as a result of random rolls, you're the only ones to blame for taking the path through this swamp and being on the move in your current conditions. Don't look at the GM, he didn't cause this.
To make this clear to the players, my approach to random encounters is to roll the random encounter check die in the open and explain to the players that an encounter will happen on a 1, regardless of whatever size of die I chose to be rolled to set the probability of an encounter. Even better, let one player roll the die. To really bring home the point that this encounter happens because of them and not of you.

The main point: If we randomize content, why do we have to make preparations?

What really is improvisation? Improvisation is finding a solution to a problem that is outside of the usual expected procedures of whatever task you're engaged in. But what people are the best at improvisation? It's the people who have a good understanding of what they are dealing with, who have a wide range of tools available to them, and who are skilled in a wide range of techniques. The best improvisation is not making up something completely new when you have nothing to work with. The best improvisation is applying the tools and skills that you have in a slightly different way than you usually do. It's very much not making shit up as you go. It's applying the tools that you came prepared with.

A random encounter table that simply lists "goblins" and "bandits" isn't really any useful for improvising great conent on the spot. But it can be useful if you have done preliminary work on what "goblin" and "bandit" specifically mean in this campaign you are currently running. The stat block for the monster in the book is just a bare bones skeleton. It's not yet a create with background, motivations, and typical behaviors. The entry of "goblins" on the encounter table for the Spider Woods can refer to a very different thing than the "goblins" entry on the encounter table in the Caverns of the Obsidian Wyrm, even though they have identical hit points, AC, damage, and saves. Doing some worldbuilding in advance on different goblin tribes in the region will go a long way in being prepared for when the party randomly encounters them. Instead of it being a group of random goblins that you have to flesh out on the spot (which you won't, so they will just be generic goblins again), you already know to at least some extend who these guys are and what they are like. That makes it much easier to give them personality and make the encounter memorable and different from other goblin encounters.

A different example of preparation is preparing wilderness lairs for random encounters in the wilderness. You roll a random encounter with 12 brigands who happen to try ambushing the party, but the players manage to defeat them without a single one getting away. Perhaps they get one alive and to talk or they follow the tracks of the brigands leading to the ambush site to find their camp. Now you need to roll up a brigand lair, which comes out as 127 brigands, six 2nd level fighters, three 4th level fighters, two 5th level fighters, one 9th level fighter, and a 10th level mage. You can roll this up pretty quickly (if you have stats for the higher level brigands prepared), but having the bandit camp simply be a big circle of tents in an open clearing would be quite boring. Especially when every brigand camp looks like that. Better preparation here means having a handful of maps for camps already ready to pull out. Have a small cave lair in a hillside, a large cave lair with wooden huts clustered along the walls, a circular pallisade with three guards towers, a ruined keep on top of a cliff, a cluster of tree houses with rope bridges between them, a small island on a lake, and whatever else you can think of. Not just as brigand camps, but also potentially goblin lairs, or gnoll lairs, or an elven hunting camp, or whatever group of randomly encountered creatures you need a lair for. You could use them as sites to encounter a pair of wyverns who are devouring the inhabitants of the camp they just killed, or a group of ghouls gnawing on the bones of warriors who lost a battle two weeks ago.

The best use for random tables is not to generate content, but to randomize content. They can help you to quickly put together elaborate encounters from previously compared components. And when the prep and improvisation is good, you can indeed generate a majority of content for a campaign. Because once a randomized encounter happens, it doesn't just have to have to end there and be done. You now can add new sites to the map that contribute to the greater tapestry that the players are exploring. If the players encounter three groups of goblins and wipe them out in the same area at different times, the players will expect there to be a lair somewhere nearby and they might want to try to find it. At that point, you don't say to yourself "they can search for as a long as they want, but they won't find anything because there is no lair". That's another point where you improvise and create a goblin lair in the area because of what has happened so far in the campaign, there really should be one. Doesn't need to be right away as the players don't need to find the lair now. If you want to make it a really cool goblin town, you can do that over the next week or two. In the meantime right now, you can let the players wander aimlessly through the forest by keeping them entertained with improvised content from your campaign's custom random tables.

 

[squeen]

I am totally on-board with randomizing content for encounters---just not-so-much with hex-contents on a map. It seems like permanent features can be pre-determined (placed) for the session, where as random encounters are best used to represent dynamic movement (i.e. things that aren't fixed in time or space) like wandering monsters or infrequent events.

As much as possible, I prefer an underlying object-permanence to the world that exists whether or not the players witness it. I think that compliments an exploratory game.

 

[Beoric]

I agree with pretty much everything @Yora , except I don't roll encounter checks in the open. I don't do this because the PCs don't always know they have had an encounter, if they fail to detect team monster and team monster decides not to engage, but to exploit their knowledge of the PCs' movements in other ways. I don't want to influence the players' decision making if they would not be aware that the situation has changed (and maybe next time they will put out scouts or otherwise try to improve their chances of detecting the enemy).

I also agree with @squeen that object permanence improves an exploratory game, when you have that option.

 

[Yora]

Monsters that can stay hidden are indeed the one big shortcoming of having the players roll the wandering monster checks.
My solution to this is to roll up a list of monsters, their number, and their surprise status for each encounter table that is likely to be relevant in the next game in advance. When wandering monsters are encountered in the areas covered by those tables, I'll go down the respective list that I made. If I know the next random encounter will be with a creature that will not be surprised, I have some time to think about what I'll do if the party gets surprised and the creature is hostile. Once the encounter does happen, the players are surprised, and the monster is hostile, I'll treat that encounter as the monster having been following them for some time and choosing this moment to strike.
It's certainly not perfect. A random encounter might happen in a place that is much worse for an ambush than the room right next to it, or it's otherwise not a good opportunity to spring the attack right at that moment. It is a tradeoff, but I think it's worth it.

Procedually generated content certain can't replace prepared stuff. It can only suppplement it. Randomly generated monster lairs have to be in addition to all the prepared stuff in the area. Once they are discovered by the players once, they'll always be in that location for the rest of the campaign. (Even if the inhabitants move out.) And it's not like such lairs appeat out of nowhere. They are the lairs of creatures that can be randomly encountered, and as such the creatures have always been there. And they would always have had a lair. It's just that the location of that lair wasn't specified until the players followed a randomly encountered creature to it, or where led there by one of them.
And when in doubt, you always have to apply the No Stupid Rule. Any roll on a random table that just doesn't make any sense in the current situation is ignored. If you encounter a group of brigands in the same hex as a town the players have been to many times without ever hearing of a brigand problem, then you wouldn't put a huge brigand camp with permanent houses right in the hex next to it. Either the lair is actually much further away from the town, or the brigands just moved their tents close to the town. Or you always have the option that the brigand group encountered just outside the town doesn't have any lair and is just passing through the area. Even if there have been lots of brigand encounters in the area before, this particular group might be from a different band and completely unrelated to the ones harrassing the area.

This becomes of course much easier to sell the larger an area the campaign covers. The bigger and more wild the environment is, the more believable it becomes that even large features have never been noticed before.

 

[EOTB]

Monsters that can stay hidden are indeed the one big shortcoming of having the players roll the wandering monster checks.

Only if the DM routinely follows through with an encounter. Sometimes the DM should recognize the other party wouldn't desire to make contact, and especially if not fantastic, wouldn't be recognized by the players as a potential "encounter-encounter". Wolves on a random table are an example of this - just because the players roll a hit on the random encounter range, if the DM rolls up "2d6 wolves" and the PCs are a troop of full-strength characters with men-at-arms tagging along tramping through the woods, it does not follow that the wolves will blindly run down from the trees and attack the party. Instead they might shadow the party for a time in case some weakling separates from the group. But whether now or shortly later, the wolves are just going to leave and seek more appropriate opportunities and the encounter will remain an unknown non-event to the players.

If even occasionally when the players roll a hit on the possibility of an encounter, nothing happens from it, eventually the tendency towards paranoia won't be default reaction. Then the rare slithering tracker, aerial servant, or what have you can still surprise players having awareness they rolled a "1" on 1d6 sometime in the recent past.

 

[Johann]

Excellent original post and insight plus good comments! Well done!

 

[Two orcs]

I agree completely with the original post. I think a the strongest game to aid your improvisation as GM with minimal (but mandated!) prep time is Apocalypse World. There is a process for breaking down all characters, locations and phenomenon into threat type, each with a certain impulse and a list of "moves" (default tactics) which makes the action-reaction part of GMing a breeze. These short lists were really the heart of that game and the reason why it was very good while most of the PbtA hacks fell flat. You described this in a less formalized way in the orginal post, thinking about what a certain goblin means, with a pre-written list like in AW you could in moments pin the Tobacco Eater Tribe as Brutes: Family or Brutes: Cult etc.

 

[Yora]

In Apocalypse World, enemies don't really have stats. And the geometry of a battlefield is irrelevant. That makes improvisation extremely easy.

 

[Two orcs]

Improvising terrain is a practiced skill which I used in both AW (where terrain does matter, though it's not supported by the rules) and D&D. In a random encounter in the wilderness, or a city, I draw the battlefield, let the player's take a position that corresponds to their direction of travel, then place the enemis at the encounter distance. Drawing buildings, hills, rocks etc. can be done on the fly both online and in person. Same with stats really, in AW you principally have attack, armor and gang size, in D&D HD, AC and numbers. For me AW was great practice for DMing D&D as it emphasizes self-bounding for the GM to keep the world impartial, make prep, let the player's loose, let the chips fall where they may.